Are you sure? - Open Source Tools for Uncertainty Enabling the Model Web

Session Type: 
Academic Session
Benjamin Pross, Institute for Geoinformatics, University of Muenster
Dan Cornford


The Model Web envisions an infrastructure for easily providing access to, and coupling of, environmental models and their results. This poses numerous conceptual and practical challenges to permit web-enablement of environmental models. One key issue is the uncertainty and trust in model results and processes that will be available to the heterogeneous users able to access the models and result via the Web. For example, when chaining services of varying or unknown quality, the uncertainty in the outputs needs to be propagated to allow rational decision making based on the final product.
The European FP7 research project UncertWeb aims to develop and apply tools and methods to create an uncertainty enabled Model Web. This presentation provides an overview of the Open Source tools developed in the project to support uncertainty management in the Model Web. This includes (i) a common data API for uncertainty enabled vector and raster data, (ii) uncertainty enabled Geoprocessing services and (iii) visualization tools for communicating the uncertainty in the model results to the users. The common data API provides a Java implementation for uncertainty enabled profiles of the Observations&Measurements model and NetCDF in various encodings by utilizing the UncertML 2.0 API. The uncertainty-enabled Geoprocessing services are also implemented in Java as profiles of OGC Web Processing Services (WPS) or plain SOAP/WSDL services. The services provide functionality for exposing uncertainty enabled models  and for transforming uncertainties in the Web. Additionally, a wrapper service, which uncertainty enables already existing WPS, is defined. The visualization tools offer different means to visualize uncertain scalar, vector and raster data in the Web. For example, time series of observations can be visualized including 90 percent confidence intervals or raster data can be visualized showing the exceedance probability for a certain threshold of interest.
The presentation is structured as follows: At first, the application scenarios of the project are introduced including biodiversity and climate change, land use response to climate and economic change, short term air quality forecasts, and individual activity modeling. The specific requirements for uncertainty support resulting from the application scenarios are discussed. The main focus of the presentation consists of an explanation of the Open Source tools as introduced above. The presentation closes with an outlook on future developments and integration with other Open Source tools.


Speaker Bio: 

Benjamin Pross is a research associate at the Institute for Geoinformatics of the University of Muenster and active developer at the 52°North Open Source Initiative. 

Schedule info